The value of subjective data, and how Google used it to build successful teams: NYT story and comment

I've argued before that subjective information is among the most important measurements of the success of emerging technology research, and that you can add subjective data into a metrics stream by assigning number values to subjective ideas.  It's a great concept, one that is not often given much weight in enterprise metrics.

I recently hurt my back at the gym, and got a referral to see an orthopedist.  Among the very first things in our discussion was the question, "how much does it hurt on a scale from one to ten?"  "Well, it's an 8 when I have a back spasm, but just sitting here right now talking with you, I'd say it's a 2 or 3," was my answer. That discussion is a practical demonstration of applying hard numbers to subjective feelings as a way to benchmark them.  It's not touchy-feely at all, and it's surprisingly practical.

image from the NYT article

And in that conversation, we have an important lesson about the way we measure performance within our enterprises, and how Google applied numeric values to subjective data to find out why some teams succeed and others don't.

After exhaustive number crunching about teams that worked really well, one of the main findings was that all team members need to have a sense of what they call "psychological safety" within the team.  In successful, high-performing teams, all members feel safe to offer any contribution without fear of ridicule or reprisal.

Here is the article:

NYT: Google research on subjective metrics and successful teams

Here is a brief excerpt:  ‘‘ By adopting the data-driven approach of Silicon Valley, Project Aristotle has encouraged emotional conversations and discussions of norms among people who might otherwise be uncomfortable talking about how they feel.  "Googlers love data,’’  (Google manager, Matt) Sakaguchi told me. But it’s not only Google that loves numbers, or Silicon Valley that shies away from emotional conversations. Most work­places do. ‘‘By putting things like empathy and sensitivity into charts and data reports, it makes them easier to talk about,’’ Sakaguchi told me. ‘‘It’s easier to talk about our feelings when we can point to a number.’

"The data helped me feel safe enough to do what I thought was right,’’  is the concluding quote in the NYT story.

Let's take this away from the idea of just talking about feelings.  As Google's research bears out, feelings turn out to have measurable value for enterprise success.  But the goal to get value-add from subjective data goes much farther.  When it comes to perceiving nuance and big picture--as opposed to big data--there is still no more powerful computer on earth than the human brain, and you have them in your organization in abundance.  Unfortunately, because it's tougher to report nuance as a hard number, it's relegated to the "subjective, hence unimportant" bin.

We already have historical evidence of subjective input outperforming choices based upon pure datasets: crowdsourced stock predictions.

Here is a practical example of measuring subjective data in business.  At each stagegate of your projects, ask participants to report their sense of the project: "on a scale of one to ten, how do you feel about this project's likelihood of success?"  A caveat, though: unless you have made giving an opinion openly a safe thing to do*, the data you receive is likely to be highly biased.  Anonymity through a third-party survey site might be in order.  Average the results, chart it over time against your other KPIs for the project, and I don't think it will take very long to get a very useful Project Health Index that you can benchmark against variables within the project management, and against other projects.  That is an early warning system, a business metric you can take to the bank.

By the way, my back is doing better.  Occasional incidents of a 5 or so, but the trend is good.



*The whole topic of "safety in reporting" is profoundly important to business metrics. I'm at work on a subsequent article on the idea, and will report steps former Ford CEO Alan Mulally took to change the fear culture that used to permeate Ford.

Why is CRM a failure?

When I first started getting paid to look at the future of technology, the very first conference I attended was about the glowing promise of Customer Relationship Management (CRM.)

It was in the spring of 2001.

Back then, CRM was as full of hype as Big Data is today.  CRM was going to give us that mythological end-to-end view of everything we knew or had done with our customers.  We would have purchase history, customer inquiries, account history and information, and all their preferences and favorite products--you name it.  If we had touched the customer or they had touched us, the CRM system would know, and would have our backs.  We would all be happy customers, enterprises would rejoice along with their newly empowered sales and customer service staff. There would be endless marketing and product information.

Put simply, it didn't happen.  It's 15 years later, and for the most part, CRM sucks.  We are a long way from "delighting the customer."

Here are some of the worst examples:

• The dreaded customer experience of trying to get help by phone.  It starts with "listen closely as options have changed," and may hit its peak when you are required to enter an account number by phone, only to finally reach a live human, who starts the conversation with "may I have your account number?"  That same human has a computer screen which rarely knows that you've bought the product you're calling about, has no clue if you've called about this issue before, or any knowledge whether you have spent $100 with them, or $10,000.
• There is a gaping disconnect between e-commerce and bricks and mortar customer knowledge.  As a customer, when have you ever experienced a company that knew your online purchase history when you walked into one of their physical locations? Or vice versa?
   
• The inability of sales staff to access useful information about the customers they are trying to serve at the point of decision.  I learned this horror story firsthand.  During my time in retail, I worked for a national upscale retailer.  They wanted--really badly--to market by e-mail to the prime existing customer base. To make that more likely, all sales staff were encouraged every day to link the customer to an e-mail address.  And since that same system was so limited in its customer search function (only by name and zip code,) the most usual result was  "I can't find you, let's just create you in the system right now."  Any associate at any POS terminal could create an entirely new customer record to add their e-mail address.
  
  The data was so littered with duplicates that it was essentially useless.  If I did succeed in finding you, it would only be one of the many versions of you. Customers' purchase histories might be under any one of a dozen versions of their name.  If they ever shopped online or at another bricks and mortar location, too bad. Each store only accessed records for the one store where we both stood, frustrated at our inability to find information that had to be in there, somewhere.
• Medical records, which are rigidly bound by rules and systems to make sure data is accurate, are astoundingly siloed and useless beyond the context in which they were created.  Your primary care physician may have a great end-to-end history of your office encounters, but if you go to a new specialist, or get admitted to an emergency room, how is that shared?  The patient hand-writes what he or she can remember.  For every single new place or practice.  

These are just the tip of the iceberg, I'm sure you have your own list based on your experience as a customer, and maybe upon your frustration at CRM efforts within your enterprise.  It's my opinion that the realization of CRM has missed the promise by such a margin that it is reasonable to say the idea has failed.  I think there are some systematic problems that make good CRM really, really hard.

• For CRM to work, you have to overcome a big problem; CRM is a genuinely vertical process, spanning almost every function of a corporation. But CRM's success depends on it working well across a range of processes, most of which are themselves internal vertical domains.  See this graphic I originally posted in 2011: most functions that form the core of a good CRM program live on the right--vertical--side. But some of the most CRM-critical are strategic horizontal processes, like marketing.
   
• Master Data Management (MDM) is excruciatingly difficult.  Talk to the data people at any enterprise and ask if they think they have it mastered.  MDM is not sexy, and it's not trendy.  The only thing is, it's just absolutely vital as a foundational element of CRM.  Circling back around to the hype on Big Data, we come to another idea: why are you worrying about big data when small data is still so underutilized, and still adding so little value?
  
• CRM succeeds or fails depending on the completeness and accuracy of information coming in as the customer does business with you.  That's not so hard during e-commerce, but in a face-to-face situation, customers have become highly averse to sharing information.  Until we find a seamless way to link the customer to their own records, and until we earn enough trust from customers for them to share willingly, CRM will remain a tough uphill slog.
• Even if you get data definitions and structure right, it is still bound to specific programs, hence badly isolated from all the necessary use cases.  If ever there was an argument that data should be offered as a service to be used in real time by different programs, CRM is it.  But that runs smack up against proprietary interests, and is not likely to change soon.  When systems for manufacturing, distribution, marketing, and point of sale are not drawing from the same data store, is it any wonder that so many enterprises have thrown up their hands and given up?
   
• There's another silo issue that is external to your business. As it stands, CRM is not personal, it is bound to specific companies.  So if it would be useful to me to know that a coffee blend at the Peet's I'm visiting is similar to one I've bought at Starbucks, that purchase history is not one Starbucks is likely to share.  If I happen to book one day with United Airlines, it would be useful to me and to United if they knew that I've flown almost a million miles with American, and am a potential high-value customer.  But AA would definitely rather my flight history remain within the confines of their own systems.
  
  
  This suggests a big business opportunity for a digital personal assistant to keep my own records of my customer history to use for my own benefit, independently of all the siloes where my purchasing life now resides.  There is a huge need for consumer-centered CRM.  Stated another way, if my customer history is going to be bound to anything, it should be attached to me.   Startup, anyone?

Along with the fact that the CRM idea of 15 years ago is so badly implemented, newer technology adds the chance for a new generation of CRM that overcomes some of these obstacles, and adds real time integration with customer activities that happen outside your enterprise--social media mentions of your product, product reviews, store check-ins, and the like.  So I still have hope, and still believe that the CRM idea is worth fighting to get right.  Could this be a need where artificial intelligence (AI) could help solve the data integration and multiple-versions-of-the-same-customer challenges?

Share your ideas here in the comments, or drop me an e-mail.  Do you know examples of companies that are getting CRM right?  Do you see other issues or opportunities I should look at in a future post?  I'm looking forward to hearing your thoughts.

HP's Cooltown: 2000 looks at the future, and predicts the Internet of Things

Infrics is going to be a little more conversational these days, so this post starts with a short story.

I became Wrigley IT's advanced technology analyst on the first work day of 2001.  In the first few years of my time there, we were deeply involved in a global single-instance deployment of SAP--at that time, the largest use of one instance of the ERP platform anywhere.  We were doing so much business with Hewlett Packard that we had a dedicated enterprise sales rep, Pete McGeehan.

Pete opened doors for me at HP to leverage the strategic relationship between the tech giant and Wrigley.  We shared information about our strategic intent, they shared information about research and product development.  I started the first of many visits to the HP Corporate Briefing center on Wolfe Road in Cupertino (the facility was torn down recently, it is the site of the new Apple spaceship building.)

It was there, probably in late 2001 or early 2002, that they showed me the Cooltown video, their look ahead at a world where "everybody and everything is connected wirelessly through the World Wide Web."  Sound familiar? It's the Internet of Things (IoT,) and here is the way they envisioned it 15 years ago:

A physical web beacon

Part of what triggered my look-back-to-a-look-forward was today's announcement from Google that Chrome for Android will now support recognition of such "web beacons."  Google calls it the "physical web." 

HP's briefing center had another feature that's come to pass.  The projectors in the conference rooms were connected to a central server, so a presenter could save a Powerpoint presentation to the server, show it from any of the conference rooms, and control it with their laptop.  It didn't have to be stored on a device that was physically plugged in to the projector, and seemed both revolutionary and sensible.

Today, I can use a Chromecast to access a presentation from my phone--either stored locally or streamed from the cloud--and display it on any monitor with an HDMI port via WiFi.  It's almost identical to the woman in Cooltown giving her sales presentation.

More of the video is now real: heads up display in cars, real-time translation of languages, personal medical monitors.  Thankfully, vending machines don't (yet) offer encouraging messages from our employers.

Finally, one more personal note.  I can't leave this post without thanking Susie Tse, who ran guest services at the HP briefing center, has become a friend on Facebook, and now works at VMWare.  And a sad note to the demolition of the center itself; it was a gorgeous modern building of white and glass, with furniture and fabrics throughout by Charles and Ray Eames, the great midcentury modern designers.  I hope Apple's new monument to itself does half as well as a beacon of style.

Embedded in retail: finding hidden value in the lowly SKU

To pay the bills while searching for my return to professional life, I did what a lot of displaced professionals have done.  I worked retail, as a sales associate.

Since this is a blog about technology and business, here is an important thing about that experience:  while I was in retail, I was also a stealth business analyst embedded in the front lines, with completely candid access to what the front line really thinks, and what they experience.


That experience led to this report.  In this case, it starts with the lowly SKU, the Stock Keeping Unit, the number next to the bar code that uniquely identifies each item you sell.

Funny thing about SKUs.  They are actually a really powerful part of the entire retail environment.  In much the same way that URLs are a window into the entire internet, the SKU is shorthand for everything you can know about the things you sell at retail.  But could it be that SKUs are really underappreciated?  As I found by working at a big national retailer, if you look at them with the right background, you might find hidden value.

I'll write more about the SKU's role in a modern retail system in another post, but here's one of the most glaring examples I found of SKU-sourced value in retail that is not being exploited:

It's the listing of SKUs in the store matched to the location of the item.  

Why does this even matter?

• People who work on the sales floor eventually learn where most SKUs are stocked, but there's a learning curve.  Experienced workers and managers spend non-value-added time showing new employees where to find things. 
• The store "map" changes every time there's a new floor set or changes in vendor-managed inventory. 
• Replenishment of stock, and restocking go-backs, can be a game of reverse scavenger hunt for retail staff, who have an item in their hand and must search for its shelf location.  
• To make it worse, the logic of stock location is often based on differing criteria. Some things are grouped by type or function, some by manufacturer, some by associated marketing campaign, some even by color.  To the restocking person on the floor, a logical placement of items with others of its same type would be a sensible thing to do. But what if "same type" is not the underlying plan, and there's no key to the intended location? 
• Staff can inadvertently tell customers items are out of stock, when in fact, they are in stock but were misplaced within your store, or have moved to a different location because of a floor set change.  



This is where your supply chain ends.
Image used under Creative Commons license

• A conservative estimate of time wasted as a result of this report's absence is not huge, but it is significant, perhaps worth $2000 a year per store. And this retailer has over 100 stores. If you could add $200K in value to your annual budget, would you think it's worth the investment of a few hours work by an IT staff member?  This does not include the lost opportunity cost of sales lost due to false stockouts or customer frustration. 

You already have this information in your marketing and inventory systems.  At one retailer where I worked, managers received new floor sets from corporate as photos of displays-to-be, with a nice tabular listing of the SKUs by shelf alongside.  So SKU location was in the system. What was not available was reverse lookup: I have the SKU, where does it belong?  This would be a really simple report to generate along with the floor set.  At its simplest, it would live as a sequential-SKU printout in a ring binder at the registers, each discrete SKU linked to its location in the current planogram and floor set mix.

Even better would be integration into the point of sale (POS) system.  Sales staff at this retailer could already scan a barcode, and the POS system would display stock information for any SKU, whether more were in the upcoming shipment from the distribution center, and even what stock was available at other stores and in the warehouse.  This is pretty basic supply chain.  But it failed to combine SKU with display location.  

To leverage the power of the SKU/location report, it could also easily be extended to the SKUs location in backstock, and for stock replenishment shipments, which one of dozens of incoming boxes contained the SKU in question.  

This latter piece of information personally cost me many sales, because there was a built in issue with the inventory report.  The store received merchandise once a week from the distribution center, in pallet after pallet of boxes containing multiple SKUs. The moment the shipment arrived on the loading dock, it was checked into inventory, but it could take days for the staff to unpack every box.  In the meantime, a SKU would show as in stock. To make a successful sale, i would have to interrupt a manager doing useful work to access a report (only accessible to managers) that linked the SKUs to the shipment box, then dig through many dozens of SKUs within the warehouse box to find the item the customer was willing to buy.  It was cumbersome, time consuming, and wasteful--so much so, that cash-in-hand sales often walked out the door from impatience.  

If you are in retail management, you know full well how much work it took to select a product, market it, get it made and shipped to your store, get a customer in the door, and help them decide to do business with you.  All of it wasted because of a SKU hiding in plain sight that lost you the sale.

There are a couple of lessons in this story.  First, I learned that there is untapped value in something as simple--yet as powerful--as a SKU.  Second, I think it's common, especially at the level of the C-suite and their direct reports, to think in terms of new solutions, rather than exploring ways to get more value out of what is already in place.  If you're a senior manager of a retail organization, you already have the resources in place to help you find that value: your own front line staff.

The front line alone deals with stock management at a granular enough level to appreciate opportunities like the SKU/location report.  But without any insight into inventory system design, or any perception that their information is valued, the value capture opportunity is lost.

Cloud Sherpas: The future of retail hinges on worker empowerment

I'm at work on the first of my reports relating lessons learned from my time as a sales associate in retail.

In the meantime, Cloud Sherpas just published this article, which has a good overview of the challenges (and opportunities!) ahead in retail.

--from cloudsherpas.com

The Future of Retail Hinges on Worker Empowerment 

Let's talk more about this soon.  For now, here are some bullet points to consider:

• The extreme personalization trend I've written about before is finding an exciting expression: the growing realization of the coming power of artificial intelligence (AI.)  Especially for lower value transactions, the day will soon arrive where an AI interface with the customer will be more satisfying to customers than one with a low-motivation, low-authority frontline staff member.  
• But...you will still have real live, look-customers-in-the-eye frontline staff. A huge opportunity awaits to empower your live front line to delight customers: with more authority to make customer service decisions, more information with which to satisfy customers, and a commensurate higher wage, reflecting the high value of interpersonal sales skills.
• In the meantime, as the Cloud Sherpas article explains, there are a lot of ways to help your current staff add more to the bottom line.  This means real time access to inventory information, a genuine full-view-of-the-customer-no-I-really-mean-it-this-time CRM system, and tight integration of your social media, marketing, and e-commerce presence with your bricks and mortar stores. In fact, these will build the foundation you need to welcome the future.

Remember, all this discussion is only about technology as a matter of peripheral interest.  The real heart of the matter is this:

Going forward, retailers, even very large ones, have the technological power to make every visitor feel as if they have stepped into a small neighborhood shop, where they're known as an individual, valued as a friend, and treated like the most important thing in that retailer's existence.   In a world of everpresent mobile access and online shopping opportunities like Amazon, that's the mandate.  What will you do to take advantage of that power, and when?

By the way, the article I'm working on will present a simple report your IT department can add to your POS system and put in sales associates' hands. It won't change your life, but I estimate it will save each of your retail locations several thousand dollars a year in lost staff time and lost sales. I know this, because I was one of those associates who ranted when I spent hours of extra work because this report didn't exist.

Coming soon.